United States Environmental Protection Agency Water Engineering Research Laboratory Cincinnati OH 45268 Research and Development EPA/600/S2-85/102 Nov. 1985 &ER& Project Summary Organic Chemical Fate Prediction in Activated Sludge Treatment Processes J. W. Blackburn, W. L. Troxler, K. N. Truong, R. P. Zink, S. C. Meckstroth, J. R. Florance, A. Groen, G. S. Sayler, R. W. Beck, R. A. Minear, A. Breen, and0. Yagi This project summary describes re- sults from a broadly based effort to determine the feasibility of predicting the fates of organic chemicals in dif- fused air. activated sludge wastewater treatment processes. The three conver- sion/removal mechanisms emphasized in this work were stripping, sorption on biomass, and bio-oxidation (biotransfor- mation and mineralization). After an initial literature review and critique, separate projects were implemented to study experimental and mathematical predictive methods on each individual fate mechanism and to develop exper- imental and/or mathematical protocols where needed. Finally, a project was implemented to couple the mechanisms in a semi-deterministic predictive equa- tion and to attempt initial verification for the equation in a continuous, com- pletely mixed laboratory activated sludge study. Specific compounds studied in this project include methyl ethyl ketone, toluene, phenol, aniline, 1,4-dichloro- benzene, 2,4-dichlorophenol, and pent- achlorophenol. Only certain compounds typical for the mechanism of interest were studied in each mechanism proj- ect. Both 14C-labeled and nonlabeled compounds were used in the sorption, bio-oxidation, and laboratory activated sludge projects to provide independent and corroborative analysis as well as an unambiguous measure of bio-oxidation mineralization. Stripping studies in a coarse bubble diffuser reactor determined the kinetic relationships between compound Hen- ry's law constants, liquid volume, and air flow rate and the effects of contam- inants on stripping kinetics. Sorption studies used nonviable biomass in a special variable-volume reactor to meas- ure sorption equilibria and to estimate sorption kinetics. Predictive equations are proposed for stripping and sorption processes. Bio-oxidation kinetics for both bio- transformation and mineralization were determined for batch, fill-and-draw, and continuous systems. First-and second- order (first-order in both substrate and biomass concentrations) kinetics were formulated. Specific degrader and total viable subpopulations were enumerat- ed. An algebraic, coupled, predictive equation and related fate equations are proposed for diffused air, completely mixed activated sludge systems. Unver- ified examples of the use of these equations for fate prediction are also presented. This Project Summary was developed by EPA's Water Engineering Research Laboratory, Cincinnati. OH, to an- nounce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction The goal of this research was to eval- uate the potential of developing a method for quantitatively predicting the fates of chemicals in an activated sludge plant—a predictive fate method (PFM). For PFM development, the theory behind the spe- cific transport and conversion mechan- ------- isms must be known and these mechan- isms must be coupled into a combined, dynamic model. Experimental data must exist or be generated on each mechanism to evaluate and verify the model. Collec- tion of these data must be controlled so that the effects of different mechanisms can be analyzed separately. The model or method must be extensively tested, ulti- mately against full-scale data. This project was conceptualized with the following objectives: 1. To survey and assess the existing data base related to stripping, bio- mass sorption, and bio-oxidation mechanism rate predictions. 2. To select a limited list of compounds for study that might be indicative of the behavior of many more organic compounds. 3. To develop experimental protocols to measure and quantify the kinetics of stripping, sorption onto biomass, and bio-oxidation. 4. To develop mathematical protocols or models to describe the removal/ conversion rates for the stripping, sorption,and bio-oxidation mechan- isms. 5. To develop an overall experimental protocol for measuring the mechan- ism kinetic rates in a continuous system more representative of real operating activated sludge plants. 6. To develop an overall mathematical protocol or model to incorporate the individual mathematical predictors and describe compound fates in a complex system indicative of an operating activated sludge plant. 7. To determine the feasibility of eval- uating the kinetic rate and other constants necessary for individual or coupled mechanism fate predic- tion from physical, chemical, or structural properties of the com- pound itself. The literature was surveyed to deter- mine the present level of understanding of the relationships between chemicals and their treatment in activated sludge treatment processes. Treatability was analyzed in terms of the basic removal/ conversion mechanisms of (1) bio-oxida- tion, (2) stripping, and (3) sorption. In some cases, other mechanisms such as chemical oxidation, hydrolysis, or photol- ysis become important. The first three mechanisms were given priority in this study. Selection criteria were developed and applied to a large list of aromatic com- 2 pounds resulting in a limited set of compounds that potentially represented a wide range of variation in each removal mechanism. Compounds ultimately studied in this project are: • Stripping: toluene, 1,4-dichloroben- zene, methyl ethyl ketone, phenol • Sorption: phenol, 1,4-dichloroben- zene, pentachlorophenol • Bio-oxidation: phenol, toluene, aniline, 2,4-dichlorophenol (very limited test- ing) • Laboratory scale activated sludge: phenol, toluene, aniline, pentachloro- phenol (biological rates minimized) A waste stream was selected that possessed time based uniformity. Pulp and paper mill foul condensate is a distillate of the liquor from the pulping process and is relatively uniform in composition over time. The focus of this study was to determine methods to predict fates of organic compounds added to this wastewater, not to determine the treatability of a specific industrial waste stream. Biologically inactive (nonviable) biomass was required for conducting experiments on the biological sorption of organics. Five techniques were tested: gamma irradiation, formaldehyde treatment, ly- ophilization, lyophilization followed by dry heating, and lyophilization followed by exposure to UV light. Biomass from the acclimation reactors was used in all experiments. Of the' methods tested, the most desirable was lyophilization of act- ivated sludge followed by dry heating at 105°C for 3 hours. This method produced a stable, dry product that upon rehydration retained the physical flocculation and settling properties of the viable biomass and appeared to have a long shelf life. Stripping, sorptive, and bio-oxidative fates of the test chemical of interest were determined for batch, fill-and-draw, and continuous laboratory-scale activated sludge (LAS) reactor conditions. Conven- tional analytical techniques using chro- matographic procedures and radiochem- ical tracers were used to determine influent and effluent chemical concentra- tions, fractions of sorbed and stripped chemicals, and biological degradation and transformation products including CO2. Stripping Predictive Fate Method Results Stripping tests were conducted with experimental equipment as shown in Figure 1. The system's liquid volume was about 26 L, and air flows from 2 to 8 L/min were used. The stripping PFM developed offers a simpler approach for estimating stripping rates in clean water and wastewater than other approaches based on the two-film theory of mass transfer. This method incorporates the Henry's law constant (Hc in torr L/g- mole) and the liquid volume-to-air flow ratio (V/Qair) to directly yield the stripping rate constant Kasta) for clean water systems. Equations predicting the strip- ping rate constant from the compound Henry's law constant follow the form of Ka sta • V/Qa,r = 3.71 x 1CT3 (Hc)1 °45 for a clean water system. Units are hr~1 for Ka sta and hr for V/Qa,,. When the dimen- sionless Henry's law constant(Hi) is used, the dimension less equation, Ka8VV/QaiI = H,, is developed. This result, along with other studies in this area, strongly sug- gest that the process of stripping from coarse bubble diffuser systems is a single-stage equilibrium process. The presence of surfactants, salts, oils, and nonviable biomass was found to vary the stripping rate, but in no case was the effect more than 50% when compared with that of clean water. The effect seemed to be the strongest for the higher volatility compounds. The values of the stripping rate con- stant, Ka sta, predicted from the model developed in this study, seem to agree reasonably well with previously reported values. This suggests the possibility of application beyond the compounds stud- ied herein. The stripping PFM was tested in con- tinuous LAS units. The measured toluene stripping rate was found to be consistent with the predicted toluene stripping rate constant. This further substantiates that the stripping PFM may be useful for prediction of compound stripping in con- tinuous "real world" systems. Biomass Sorption Predictive Fate Method Results Sorption experiments were conducted to determine sorption kinetic rate con- stants and equilibrium relationships be- tween aqueous phase concentration and loadings onto biological solids for three test compounds. Test compounds were chosen to span a wide range in potential for sorption as indicated by the octanol- water partition coefficient (Kow). Com- pounds that were studied include phenol (Kow = 31), 1,4-dichlorobenzene (Kow = 2455), and pentachlorophenol (associated form, Kow = 132,000; dissociated form at pH = 7, Kow = 7,700). All octanol-water ------- partition coefficients in this summary are unitless ratios of equilibrium compound concentrations in the octanol and water phases. Tests with phenol and 1,4-dichloro- benzene indicate that equilibrium was reached in less than 15 minutes. Penta- chlorophenolsorption appears to follow a two-step process, with a very rapid initial sorption step and subsequent slower approach to equilibrium. For batch testing where sorption is the sole removal mechanism, the equilibrium aqueous concentration and substrate loading on biomass may be expressed: Cae = (K0*XfL/1000pL)+1 C, = where: Cae - equilibrium aqueous phase sub- strate concentration (//g/L) Cs = solids loading on biomass (yug/g) Cao = initial aqueous phase substrate concentration (//g/L) Kow = unitless concentration ratio octa- nol/water partition coefficient X = solids (biomass) concentration (mg/L, dry basis) fL = lipid weight fraction of biomass PL = lipid density (g/L) Table 1 presents a summary of the measured and predicted endpoint (equil- ibrium) values of Cae for the sorption experiments. Table 2 presents measured and predicted percent removals calcu- lated from data in Table 1 and the starting concentrations. Table 3 presents meas- ured and predicted loadings (Cs). Generally, good agreement is seen between measured and predicted values. Exceptions occur for experiments SP-1 and 2, the start-up experiments, and SP-6 where the mixing intensity was reduced and the experiment was run at 4°C. Bio-oxidation Kinetic Measurements The contribution of bio-oxidation to predictive fate assessment of organic pollutants in industrial waste treatment was determined for batch, fill-and-draw, and continuous LAS bioreactors. Bio- oxidation of specific pollutants was meas- Wat er In To Gas Nitrogen In Sampling System (Figures 5, 6, and 7) Symbols PR—Pressure Reducer R—Regulator FE—Rotameter TE— Thermometer D—Drain SP—Sample Port PSV—Pressure Relief Valve DO—Dissolved Oxygen Probe PI—Pressure Indicator Air In Dew Point Hygrometer Manometer Figure 1. Apparatus Used in the Stripping Experiments ured in activated sludge samples under batch assay conditions. Mineralization (oxidation to C02) of 14C-labeled pollutants was chosen as a primary and unambig- uous measurement of bio-oxidation. Qualitative and quantitative measure- ments of specific degradative bacterial populations comprising the activated sludge were performed. These measure- ments and associated enzymatic activites were used to evaluate the potential for describing bio-oxidation kinetics as a function of the biological community. In addition, the potential for predicting qual- itative and quantitative pollutant fate relative to sludge composition and activity was investigated. Laboratory-Scale Activated Sludge Study The purpose of this study was to quantify the removal mechanisms of four organic substrates in continuous flow, completely mixed activated sludge units. Two 11-L LAS units were operated at mean cell residence times of approxi- mately 5 and 10 days. These data were to be used to test the predictive fate methods that had previously been developed for stripping, sorption, and bio-oxidation. Material balances around the activated sludge units revealed that the fate of phenol and aniline are almost identical, with >99.8% of the parent compound bio- ------- Table 1. Comparison of Measured and Predicted Sorption Endpoint Concentrations for Batch Experiments Measured Predicted Concentration Concentration' Percent" Compound Experiment (mg/L) fmg/L) Deviation Phenol SP-3 SP-4 SP-5 SP-6 1,4-Dichlorobenzene SP-1 SP-2 SP-10 SP-11 Pentachlorophenof SP-7 SP-8 SP-9 SP-11 73.0 73.3 71.8 76.8 26.6 54.2 13.9 2.7 0.8 1.5 0.3 0.7 74.2 74.6 72.8 75.7 17.8 33.8 14.3 3.5 0.91 1.22 0.78 0.99 -1.6 -1.8 -1.4 1.4 33.0 37.6 2.9 -29.6 -13.8 18.7 -160 -41.4 "Where fL - 0.2 and PL = 900 g/L. "Percent deviation = (measured value - predicted value) -f- measured value x 100. cPentachlorophenol partially ionized at pH = 7, Kow = 7700. Table 2. Comparison of Measured and Predicted Sorption Percent Removal Compound Phenol 1 ,4 -Dichlorobenze ne Pentachlorophenof Experiment SP-3 SP-4 SP-S SP-6 SP-1 SP-2 SP-10 SP-11 SP-7 SP-8 SP-9 SP-11 Measured Removal f%) 4.6 5.8 3.9 1.3 43.0 30.0 72.7 78.7 86.9 76.9 97.2 92.2 Predicted Removal* (%) 3.0 4.1 2.5 2.7 61.9 56.4 72.0 72.4 85.1 81.2 92.7 89.0 Difference" (%) 1.6 1.7 1.4 -1.4 -18.9 -26.4 0.7 6.3 1.8 -4.3 4.5 3.2 "(initial concentration - endpoint concentration) 4- initial concentration x 100. "Difference = measured removal - predicted removal. cPentachlorophenol partially ionized at pH = 7. /fow = 77OO. Table 3. Comparison of Measured and Predicted Loading for Sorption Batch Experiments Compound Experiment Phenol SP-3 SP-4 SP-5 SP-6 1,4 -Dichlorobenzene SP- 1 SP-2 SP-10 SP-11 PentachlorophenoP SP-7 SP-8 SP-9 SP-11 Measured Loading (mg/g) 0.8 0.7 0.8 0.3 6.7 9.8 7.9 2.7 1.6 2.0 1.4 1.7 Predicted Loading' (mg/g) 0.5 0.5 0.5 0.5 9.7 18.4 7.8 1.9 1.6 2.1 1.3 1.7 Percent Deviation 37.5 28.6 37.5 -66.7 -44.8 -87.8 1.3 29.6 1.9 -4.0 5.0 0.6 *C, Values "Pentachlorophenol partially ionized at pH = 7, /Cow = 7700. oxidized and stripping losses below de- tectable limits. Approximately 56% of the feed 14C phenol was recovered as 14COz for both compounds. Removal of these compounds from activated sludge units by sorption is insignificant on a material balance basis, with average compound loadings of <84 /ug/g MLSS and 28 /ug/g MLSS determined for phenol and aniline, respectively. Biological oxidation was also deter- mined to be the major removal mechan- ism for the toluene, the most volatile of the test compounds. Approximately 67% to 70% of the feed toluene 14C was recovered as 14CO2 and only 5% to 6% of the 14C was recovered as soluble metabo- lites. The ability of the activated sludge systems to biologically degrade toluene was found to be very sensitive to devia- tions from steady-state operating condi- tions (flow fluctuations, etc.). Grab sam- ples of mixed liquor indicated variations in the concentration of toluene in the aqueous phase to range from <0.05 mg/L to 4.7 mg/L. Pentachlorophenol was found to be extremely resistantto biological oxidation and stripping. Sorption onto sludge was determined to be the major removal mechanism, with both 14C analyses and Pentachlorophenol analysis indicating that 6% to 8% of the feed pentachloro- phenol was removed through sorption to waste solids. The average pentachloro- phenol loading on biological solids was determined to be 1967 pg/g MLSS and 1801 ug/g MLSS for reactors 1 and 2, respectively. The predictive fate method, which was developed for stripping, was determined to be accurate in predicting the concen- tration of test compounds in the vent gas as a function of the measured concen- tration in the aqueous phase, the Henry's law constant, and the relative air flow/ liquid volume ratio. Vent gas concentra- tions of toluene predicted from daily average data were within ±50% of ob- served values for approximately 70% of the observations. The remaining observa- tions had either or both vent gas and/or mixed liquor concentrations below the detection limit. Average loading of aniline, phenol, and Pentachlorophenol onto biological solids were found to be related to the octanol- water partition coefficient. However, the proposed predictive equation for relating equilibrium loading to the equilibrium aqueous phase concentration was found to predict loadings that were lower than those observed for aniline, an ionizable ------- compound. Conversely, predicted penta- chlorophenol loadings were seven-fold higher than those observed. Overall Mathematical Predictive Fate Method An overall mathematical PFM was developed to predict the equilibrium aqueous phase concentration in the secondary effluent from a completely mixed activated sludge system. The strip- ping PFM and the equilibrium sorption PFM for substrates at low concentrations are incorporated into traditional "uniform biomass" design models taken from environmental engineering. This equa- tion, and its derivation, is given in the project report. Conclusions This work represents a broad overview assessing the feasibility of using both experimental and mathematical predic- tors to quantitatively determine fates. Methodology using chemical engineering kinetic approaches and reactor-design equations appears to be a viable way to analyze multi-mechanism processes for various reactor configurations. The data base existing at the beginning of this effort was limited because of the specific emphasis given to different ques- tions in each study and to the past focus on removal as opposed to specific chem- ical fates. Some indications on relative fate processes may still be drawn, partic- ularly from more recent studies that look at numerous compounds with the same experimental protocol. Stripping in completely mixed, diffused air systems has been found to be essen- tially an equilibrium process, and predic- tive equations are proposed. Verification on other compounds and system config- urations is suggested. The effects of contaminants(surfactants, salts, oils, and nonviable biomass) are significant on oxygen transfer rate constants and less significant on stripping transfer process- es. Stripping rate estimates based on oxygen transfer rates for contaminated waters such as treatment plant mixed liquors must be questioned. Sorption was quantified in a variable- volume batch reactor using nonviable biomass and later in continuous labor- atory activated sludge studies. An equil- ibrium deterministic predictive equation is proposed for low aqueous concentra- tions. Kinetic sorption rates for the com- pounds studied could not be quantified because of their rapidity compared "with the time required for sampling and anal- ysis. Numerous kinetic rate constants for biological processes exist and can be formulated. Each of these rates applies to specific cases in substrate and biomass concentration and reactor configuration. Little reliable data exist in this area, and accepted overall methodologies are lack- ing. First- and second-order (first order in both substrate and biomass concentra- tion) rate constants are developed in this study for batch, fill-and-draw, and contin- uous reactor configurations for the com- pounds studied. Disappearance rates were calculated, but mineralization rates were emphasized. Mineralization pro- vides an unambiguous predictor of bio- degradation, but disappearance kinetics are required for fate predictive equations. Mineralization rates for the compounds studied were generally comparable over the batch, fill-and-draw, and continuous reactor configurations. However, disap- pearance rate kinetics were comparable only for batch and fill-and-draw systems and ranged from 2 to 3 orders of magni- tude slower than those demonstrated in the steady-state continuous systems. The batch and fill-and-draw kinetic disap- pearance values are of the same order as data reported in the literature for specific compounds and BOD. Laboratory-scale activated sludge sys- tems were designed and operated to elucidate compound fates in a conclusive fashion. This included the use of radio- labeled substrate in conjunction with nonlabeled substrate and sampling of all streams of fate importance. Many of these procedures are substantially more diffi- cult to implement in full-scale processes. Full-scale fate data cannot, therefore, be as precise or conclusive as the more controlled experiments. A coupled, algebraic, predictive fate equation is presented subject to assump- tions and derivations given in the project report. An example of organic compound fates is provided by assuming most probable, maximum, and minimum values for the equation variables and using biological rate constants calculated from other studies. Generally, the most prob- able values agree with the findings in this study, and the ranges agree with full- scale plant studies implemented by EPA. The full report was submitted in fulfill- ment of Contracts No. 68-03-3027 and 68-03-3074 by IT Corporation and the University of Tennessee (subcontract) under the sponsorship of the U.S. Envi- ronmental Protection Agency. U. S. GOVERNMENT PRINTING OfFICE: 1985/646-116/20718 ------- J. W. Blackburn, W. L. Troxler, K. N. Truong, R. P. Zink. S. C. Meckstroth, J. R. Florence, and A. Groen are with IT Corporation, Knoxville, TN 37923; G. S. Sayler, R. W. Beck, R. A. Mi near, and A. Breen are with University of Tennessee, Knoxville, TN 37916; and O. Yagi is with National Institute for Environmental Studies, Yatabe, Japan. R. J. Turner is the EPA Project Officer (see below). The complete report, entitled "Organic Chemical Fate Prediction in Activated Sludge Treatment Processes," (Order No. PB 85-247 674; Cost: $28.95, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Water Engineering Research Laboratory U.S. Environmental Protection Agency Cincinnati, OH 45268 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati OH 45268 Official Business Penalty for Private Use S300 EPA/600/S2-85/102 0000329 PS U S FNVIR PROTECTION AGENCY REGION 5 LI^ASY 230 S DEAR30RN STREET CHICAGO U- 60604 ------- |